In terrestrial ecosystems, the rhizosphere, which is the zone of adherence of soil to plant roots, is one of these hotspots because the biomass and activity of microbes are enhanced as a result of exudation from roots of compounds such as organic acids, sugars, amino acids and vitamins [17]. The genera Agrobacterium [18], Burkholderia [19], Enterobacter and/or Pantoea [20], Herbaspirillum [21,22,23], Ochrobactrum [24], Pseudomonas [25] and Stenotrophomonas [26], among others, entertain bivalent interactions with both plants and human tissues. Except for the notorious phytopathogen Agrobacterium [18] and the opportunistic plant pathogen P. aeruginosa [25], most of these rhizobacteria promote plant growth by various mechanisms, including antagonism against phytopathogens.

The fact that many bacteria in the rhizosphere are antibiotic producers could explain the frequent detection of bacteria with multiple antibiotic resistances in this micro-environment [17]. Moreover, several members of these genera show a common ability to degrade environmental pollutants [27,28]. This particular capacity to degrade xenobiotics is related to an extensive enzymatic arsenal that could play a role in the resistance to both cellular toxic products, such as free radicals [29] and antibiotics [30] observed in the clinical strains. In addition to xenobiotic-degradation enzymes, transporter pumps that push out xenobiotic molecules, including antibiotics, from inside bacterial cells have been widely detected in bacterial genomes and particularly in plant-associated bacteria [28]. Transporter overexpression, due to mutations in regulatory genes, can be easily selected using antibiotics and antiseptics, but also chemical compounds used for housekeeping, farming, and food processing [31]. Recently, two-component signal transduction systems were shown to control the expression of xenobiotic transporters that are transiently induced by environmental stimuli, such as low pH and osmotic changes, as encountered in both environment and infection sites [31].

The importance of xenobiotic transporters revealed by genomics and post-genomics suggests that xenobiotic resistance is not their natural function but that they probably have physiological substrates to transport [32]. Therefore, besides their natural function, versatile secondary functions, including the efflux of xenobiotics, became a ‘welcomed’ side effect. It has been hypothesized that bacterial virulence could be a by-product of xenobiotic transportation. Although this has not been studied in rhizosphere bacteria, recent studies on Salmonella enterica have shown that the xenobiotic transporter not only confers antimicrobial resistance but also contributes to virulence [33]. Reasons evoked are the capacity to transport secreted proteins necessary to establish virulence, such as toxins, the capacity to efflux antibacterial substances produced by the host and the capacity to transport factors effecting bacterial regulatory functions within the host, like signal transduction substances [31].

In addition to xenobiotic transporters, the lifestyle of bacteria in the rhizosphere predisposes them to become OBPs, thanks to several properties: production of bio-surfactants, competition for nutrients and minerals and degradation of pathogenicity factors naturally produced by phytopathogenic bacteria, which could be similar to those produced by the human immune cells. It has been suggested that plant response to bacterial toxins, flagella and lipopolysaccharide (LPS) resembles in many ways mammalian innate immunity [17]. Consequently, interactions with plant roots might pave the way for bacterial adaptation to mammalian and human cells.

Amoebae normally feed on bacteria, but Rowbotham in an initial report on Legionnaires' disease noted the capacity of Legionella pneumophila to survive and multiply within amoeba. This finding was linked to the fact that macrophages, the natural human targets of legionellae, are indeed amoeboid cells, despite their evolutionary distance from amoeba [34]. It was then suggested that adaptation to amoebae served as a pre-adaptation stage to the macrophage internal environment, an important step in the process of becoming a human pathogen. Since these observations, molecular evidence has shown the similarities in the intracellular infections of macrophages and protozoa by L. pneumophila as well as similarities in the lifecycles of the bacteria within amoeba and macrophages [15]. Amoebae are now known to serve as a reservoir for various other pathogenic bacteria, including the OBPs Escherichia coli, Burkholderia cepacia, Ralstonia pickettii, Listeria monocytogenes, Mycobacterium spp. and P. aeruginosa [15]. Moreover, the amoeba operates as a ‘Trojan horse’, introducing the pre-adapted bacterium into the human host [35,36]. Another teleological metaphor is proposed by Schmitz-Esser et al. [37]: amoeba serves as a ‘training ground’ for bacteria that would acquire the capacity to live in a human organism, should they meet one. This strategy has also been observed for Mycobacterium avium, an opportunistic environmental pathogen for children and patients with AIDS; the co-culture of amoebae with M.avium enhanced the latter’s virulence, particularly its motility and ability to cross the murine intestinal epithelium [38]. Moreover, an enhanced resistance to antimicrobial agents, to biocides and to cold was observed [36]. The resistance of the amoeba-associated bacteria to cold suggests that amoebae serve as bacterial reservoirs at low temperatures in the environment but also on human body surfaces, such as nasal mucosa; subsequent lysis of the amoeba at higher temperature results in the dispersal of the bacteria, for example in the lower respiratory tract [15]. Virulence enhanced by amoebae is not limited to intra-amoebal or intracellular bacteria, amoeba-bacterium interactions in extracellular contexts occur as well. For instance, Acanthamoeba castellanii interacts with Vibrio parahaemolyticus without any engulfment, but rather by secreting a substance that promotes the survival of the bacterium in co-culture [39]. Free-living amoebae have also been described as melting pots for bacterial evolution, fostering horizontal genetic transfers (HGTs). For example, L. pneumophila and Protochlamydia amoebophila, which both live in amoebae, share genes with a phylogenetically unrelated intracellular bacterium, Rickettsia bellii [13]. This suggests the co-existence of their ancestors in amoebae prior to diverging evolution toward different life-styles, in amoeba and occasionally in human cells for L. pneumophila, in tick cells only for R. bellii and in amoebae only for P. amoebophila.

Humans live in close association with a variety of insects and other arthropods, which are well known as vectors for human diseases, such as malaria, leishmaniasis and arbovirus infections. From a phylogenetic point of view, Bacillus anthracis and Bacillus thuringensis, devastating for mammals and insects respectively, are believed to have evolved from an insect-associated Bacillus cereus-like ancestor that belonged to so-called "pre-vertebrate" pathosphere [14]. Common mechanisms are shared by pathogenic bacteria in the infective processes against both human and insect hosts. These mechanisms include adhesion to the host, entry and invasion of the host, establishment, dissemination within the host, toxin production, avoidance of host immune responses and transmission. At the genomic level, pathogenicity islands from mammalian and insect pathogens show astonishing similarity, suggesting that there is conservation of the genes involved in human and insect pathogenesis [14]. Moreover, human and insect immunity are also similar in many ways [16]. The analogies between insect and mammalian immunities, as well as between insect- and mammalian-associated bacteria virulence traits, suggest insects may act as a host reservoir of pathogens pre-adapted to the human innate immune system and who may ‘jump’ to humans, creating novel diseases [16].

The example of Acetic Acid Bacteria (AAB) illustrates this fact. AAB oxidize alcohols or sugars, leading to the production of acetic acid. They are commonly associated with plants and have been used in industrial food processing throughout human history, especially to convert wine to vinegar and to produce tropical fermented products [40]. It has recently been found that AAB of the genera Acetobacter, Gluconacetobacter, Gluconobacter, Asaia, Saccharibacter and Commensalibacter establish natural symbiotic relationships with the midgut of flying insects such as fruit flies, mosquitoes, honey-bees and leafhoppers [41]. This midgut niche has an acidic pH, a selective factor that supports AAB growth. In addition, AAB possess the capacity to migrate from midgut to other organs such as reproductive organs or salivary glands and hence be transmitted by vertical and horizontal routes. They also play a major role in the regulation of the insect innate immune system. Taken together, the data indicate that AAB represent novel secondary symbionts of insects [41]. In parallel, AAB are emerging as human OBPs. The first report of human infection involving AAB was a case of peritonitis associated with Asaia bogorensis [42]. Since then, AAB have increasingly been reported as responsible for human infections, particularly in CF patients [43]. The ability to grow in acidic conditions, as observed in insect midgut, may confer on AAB a selective advantage in the abnormally acidified airways of CF patients [43]. Besides this common adaptation trait, analogies between insect and mammalian immunity and susceptibility to bacterial virulence factors as well as the intimate association of AAB with both humans (through food) and insects (through plants) might drive the evolution of these environmental bacteria toward becoming emerging insect symbionts and human OBPs.

Some insect-associated bacteria can adapt to a larger range of hosts. Serratia marcescens displays adaptation to insects and plants but also to human beings. The jump from insects to plants has been observed for phytopathogenic bacteria. Intimate associations with plants predispose phytopathogens to frequent encounters with herbivorous insects that could evolve to become specific vectors for pathogens or alternative primary hosts for phytopathogenic bacteria [44]. S. marcescens is a phloem-resident pathogen that causes yellow vine disease of pumpkins and squashes transmitted by the squash bug, Anasa tristis [45]. Not only adapted to plants and insects, S. marcescens is also an emerging human pathogen that is predominantly involved in HAIs, particularly in neonatal intensive care units [46] and also implicated in a range of ocular infections [47]. One factor that may enable S. marcescens to thrive in a variety of different environments is its ability to colonize surfaces and form biofilms. Its multiple processes of surface colonization are quorum sensing-regulated, and include swarming motility, biofilm maturation and detachment in the environment. These quorum sensing-regulated properties allow S. marcescens to become a good colonizer of living surfaces in plants, insects and humans as well as on inanimate surfaces such as medical devices or corneal lenses [48]. Adaptation to one host could allow for pre-adaptation to another one, hence producing specialization in each host while widening the host range.

We will show below in detail (see chapter below: Horizontal Genetic Transfers give evolution a boost, example of Photorhabdus luminescens) that worms as well as insects and other invertebrates can be considered as shelters or ‘nurseries’ in which adaptation to the mammalian niches is prepared. Worms clearly belong to the so-called ‘pre-vertebrate pathosphere’.

The successful use of Caenorhabditis elegans as an infection model to understand microbe-host interactions and to study pathogenesis shows that common mechanisms of virulence are shared among the different kingdoms of life. Today, approximately 40 bacterial and fungal pathogens are known to harm C. elegans. The regulatory networks of bacterial pathogens seem to be conserved across multiple infection models suggesting that they co-evolved during the history of interactions between Bacteria and Eucarya. As a consequence, some of the bacterial regulators, such as quorum sensing molecules and two-component systems, interact with innate immune functions shared at the interkingdom level. The C. elegans model is particularly efficient in detecting virulence traits in OBPs because virulence in strict pathogens is more often host-specific and cannot be universally modeled in the nematode infection system [49].

Some species of environmental bacteria have an extraordinary range of behaviors and activities, such as free life in water, breakdown of pollutants, production of original biomolecules, beneficial effects for plant growth and health, phytopathogenic effects, association with animals and also multi-drug resistance and human pathogenicity. The wide range of activities of versatile life-style bacteria involves adaptation to various niches, which is supported by a larger genome than phylogenetically related specialized bacteria [52]. The content of 115 bacterial and archeal genomes analyzed by Konstatinidis and Tiedje [53] showed that larger genomes (>6 Mb) accumulate secondary metabolism, and, to a smaller degree, energy conversion-related genes. They also display an obvious increase in regulatory genes to successfully control the extensive metabolic repertoire they express under different growth conditions. In the Konstatinidis and Tiedje study [53], all species with genome size >6 Mb lived in very diverse environments with a great range of substrates for energy production. In particular, free life in water or soil provides the bacterium with varied but scarce sources of nutriments [53]. Bacteria with large genomes are frequent in nature. In spite of the bias induced by preferentially sequencing bacteria with small genomes, which are also frequently pathogenic for humans [54], large genomes of more than 6 Mb are widely represented. Consequently, one may consider that such genomes are ecologically successful with little penalty for the slow growth generally related to high genome size [53,55].

Chromobacterium violaceum is a free-living microorganism that populates the soil and water in tropical areas around the world [56], and is abundant in the Rio Negro in Amazonia. C. violaceum is mainly a saprophyte but is also an OBP for infants and immunocompromised patients. However, the fact that the Rio Negro is the source of drinking water for the population living around it, without there being widespread infection, indicates the low infectivity of this organism. While C. violaceum infections are scarce, their fatality rate is high [29]. Total genome sequencing of C. violacaeum reveals that its adaptability and versatility depend on genes specifically related to interaction with and response to environmental variations. Most represented among encoded proteins are those involved in versatile pathways for energy generation, transport, stress adaptation, motility systems and quorum sensing for control of inducible systems. Of note: C. violaceum produces an exceptional range of enzymes including paraquat-inducible proteins, drug and heavy-metal resistance proteins, multiple chitinases, and proteins for the detoxification of xenobiotics [56]. Violacein is the major secondary metabolite of C. violaceum and responsible for the pigmentation of the bacterium. The violacein pigment, which has already been used as a therapeutic compound for dermatological purposes, also exhibits antimicrobial activity against Mycobacterium tuberculosis, Trypanosoma cruzi, and Leishmania sp. and is reported to have other bactericidal, antiviral, and anticancer properties [56]. The Brazilian consortium credited with sequencing the genome of C. violaceum expresses hopes that their work will not only benefit biotechnological and pharmaceutical industries in the developing world, but will also provide “a further stimulus to the preservation of the precious ecosystems where these organisms are found.” The realization that a bacterium can belong to our biodiversity heritage is infrequent enough to be noted.

Regarding pathogenicity in humans, type III secretory system (T3SS) components similar to those in Salmonella enterica serovar Typhimurium and Yersinia pestis have been found under an incomplete form in C. violaceum. The similarity of this T3SS with those of human pathogens suggests that it contributes to human infection, its incompletion explaining that human infections are very infrequent and limited to predisposed patients [56]. Besides T3SSs, which are typical and specific virulence factors for humans and animals, several other genes are believed to be involved in pathogenicity. C. violaceum possesses non-specific virulence factors such as pili, flagella, LPS and peptidoglycan [29,56]. However, most bacteria display these factors without being pathogenic. Rather than true pathogenic traits, they could be considered as versatile factors that also support the adaptation of C. violaceum to its environmental niches. The closest similarity between the corresponding genes and their homologs in pathogenic bacteria was found for Neisseria meningitidis and P. aeruginosa. This has been evoked to explain the pathogenicity of C. violaceum [29], but is not fully convincing because N. meningitidis and P. aeruginosa are phylogenetically Gammaproteobacteria, related to C. violaceum. This relatedness alone could explain the high similarity level obtained with the corresponding genes in N. meningitidis and P. aeruginosa. The relationships between environmental specific functions and pathogenesis are better illustrated by the detoxifying capacities of C. violaceum. A part of the arsenal for detoxification of environmental xenobiotics revealed in C. violaceum genome contributes to reduce harmful free radicals. This function could be considered as an environmental pre-adaptation for bacterial evasion of the human host immune system by resistance to the free radicals produced by phagocytes. Clinical isolates show a 30% higher superoxide dismutase activity and a five-fold higher catalase activity than the corresponding activities observed in soil strains [29]. This observation suggests that strains exhibiting a powerful reduction of free radicals are successful in the human niche. As another example, violacein presents different properties (see above), all linked to its generalized cytotoxic capacity [29], which could play a part in the infectious process.

The genome size of C. violaceum is about 4.7 Mb, which is not so large considering its metabolic versatility. This is also true for the OBP Ochrobactrum anthropi, whose genome size varies widely from one strain to another among the 9 studied [57]. The genome size variability within a species suggests that the functional versatility observed for a given strain is amplified at the population level. For these bacteria, the pangenome reflects the variability of the flexible genome and the overall coding capacities of the species.

Besides the size, genomic fluidity resulting in rearrangement events is another way to increase variability, and thus improve versatility, at the population level. Ogier et al. [58] carried out an interesting study of comparative genomics on Photorhabdus luminescens variants obtained from their natural hosts. They followed short-term genomic rearrangements during genome speciation, thereby showing the role of intragenomic rearrangements in the processes responsible for bacterial genome diversification and evolution. Due to the modular organization of the chromosome, rearrangements lead to a combinatorial process rather than to a random patchwork. The fluid nature of genomes can be seen as ‘natural genetic engineering systems’ that facilitate the evolutionary rewriting of information [51]. In particular, rearrangements maximize the chances of success by using combinatorial processes based on basal functions while maintaining a reasonable genome size [51].

Genome fluidity in a species is easily detected by the use of DNA macrorestriction methods. For instance, RFLP followed by Pulsed Field Gel Electrophoresis (PFGE) shows highly diverse migration profiles for genomes subjected to extensive rearrangements [59]. PFGE is currently one of the most performant methods in molecular epidemiology of infectious diseases, particularly when infections are caused by P. aeruginosa, enterobacteria or other OBPs [59,60]. In particular, PFGE is more discriminative than sequence-based methods such as multilocus typing, which indicates that genome rearrangement is the principal and/or the most rapid mode of evolution in these bacteria.

In summary, fluidity by intragenomic rearrangement is a major process of diversification and evolution of the OBPs genomes. Another cause of genome fluidity in bacteria is the acquisition of foreign DNA by horizontal genetic transfer (HGT).

Large bacterial genome size has been related to the population size and the rate of HGT [61]. It has been shown that free-living species of large population size accumulate insertion/deletion and rearrangements at much higher frequencies than host-dependant bacteria that encounter a bottleneck in population size. In the latter, the influence of HGT is negligible and evolution occurs mainly by nucleotide substitution [62].

Expansions and contractions in the genomic repertoire mainly affect genes involved in environmental interactions. Duplications and HGTs allow a gain of information more rapidly than punctual mutations. These mechanisms of high-rate evolution enable rapid responses to alterations in the environmental conditions subjecting the bacterium to strong selective pressure [61]. If environmental conditions vary frequently, acquired characters will be conserved in large genomes, expanding and diversifying the metabolic and regulatory capacities of the bacterial cell.

Comparing the genomes of closely related strains or species with different lifestyles could help to understand the role of HGT in the emergence of OBPs. The genus Photorhabdus provides a valuable example [63]. P. luminescens, the only terrestrial bacterium exhibiting bioluminescence, lives in symbiosis with soil entomopathogenic nematodes. This bipartite natural system is currently used for the biological control of crop pests. Some Photorhabdus strains belonging to the species Photorhabdus asymbiotica provoke invasive soft tissue infections and bacteremia in humans. P. asymbiotica has a smaller genome than that of the insect pathogen. The difference in size is about 600 kb, which is to say about 600 genes and 10% of the total genome size. Moreover, one megabase of DNA appears unique to each strain sequenced [63]. Contrasting with its smaller genome, P. asymbiotica is the only Photorhabdus to carry plasmids [63]. The plasmid pAU1 reveals a wide array of transposons similar to those found in genome and plasmids of Y. pestis. The genomes of P. asymbiotica and P. luminescens show strong synteny across much of their length. However, there are several large-scale inversions in the central regions of the chromosomes associated with numerous transposons and repeat sequences. The genome of P. luminescens has been previously shown as highly plastic with genomic variations occurring in clonal populations. The phenotypic consequences of these genomic changes are cryptic but a role in adaptation to environmental conditions is suggested [64].

Differences in gene content between P. luminescens and P. asymbiotica consist in several genes coding for human host-specific virulence factors, notably genes coding for T3SSs. Effectors of T3SS that inhibit phagocytosis of P. luminescens following its translocation into insect hemocytes were replaced in P. asymbiotica by the ExoU-like effector. ExoU is a P. aeruginosa toxin with a phospholipase activity that disrupts human macrophages and has also been implicated in the T3SS-mediated killing of amoebae that graze P. aeruginosa biofilms [65]. It is noteworthy that a human pathogen-associated virulence factor, namely a cell invasion factor translocated via the T3SS, was also detected in P. asymbiotica. This gene, named sopB, is important in ‘directing traffic’ in the early stages of Salmonella enterica serovar Typhimurium entry into host cells by modulating interaction of Salmonella-containing vacuoles with the endocytic system [63,66].

Traits that enable P. asymbiotica to resist the fast-acting human innate immune response are also detected. Indeed, a small protein with homology to the attachment invasion locus protein Ail, that gives Yersinia pestis resistance to human complement, is secreted at 37 °C by P. asymbiotica but not at 30 °C [63]. Other genes found in P. asymbiotica, but not in other Photorhabdus strains, display homologies with various genes of human pathogens such as Vibrio cholerae, V. parahaemolyticus, Y. pestis, P. aeruginosa and Bordetella parapertussis [66]. Comparative genomics between P. luminescens and Yersinia enterocolitica identified several common loci representing ancestral clusters of genes important in pathogenesis. These might have evolved during the association of the two bacteria with invertebrates and then adapted to more recent pathologies in mammals. Examples are yersiniabactin, quorum sensing-like regulators, or the urease operon [67]. Y. enterocolitica, similar to Y. pestis, which has developed a strategy to infect and proliferate in insects. Y. pestis probably met P. asymbiotica in the ‘pre-vertebrate pathosphere’ as Y. enterocolitica met P. luminescens. This suggests that the emergence of P. asymbiotica as an OBP is mainly due to HGT from other human pathogens. The emergence of P. asymbiotica as a human pathogen is an emblematic example that reinforces the hypothesis of an environmental ‘nursery’ for human OBPs, detailed above: insect and nematode immune systems as well as potential relationships with amoeba make the bacterium able to evolve from resisting invertebrate immune system and/or amoeba grazing to resisting the human immune system by taking refuge in macrophages in the early stages of infection.

HGT events are obvious in most OBPs but are probably not sufficient to spawn the emergence of OBP behavior as exemplified by the case of Burkholderia cenocepacia. The strain J2315 of B. cenocepacia belongs to an epidemic lineage named Edinburgh–Toronto (ET-12) associated to CF patients whereas other strains of the species are more versatile and are found in soil, plant and man. Its 8.06-Mb genome comprises three circular chromosomes and a plasmid and encodes a broad array of functions typical of this metabolically versatile genus, as well as numerous virulence and drug resistance functions. Comparative analysis revealed that 21% of its genome is unique and highly specialized in comparison to other strains of B. cenocepacia [68]. Among J2315 specific genes, a new collagen-binding trimeric autotransporter adhesin with no bacterial orthologs was found. This gene has a role in cellular adhesion and virulence [69] and was acquired by HGT, although the parental donor remains unknown.

HGTs undoubtedly promote and contribute to the success of environmental bacteria as OBPs but other signs of genomic evolution were detected in the J2315 strain. For instance, pseudogenes indicative of genomic reduction are detected, suggesting that both gain and loss of functions have occurred in this OBP, specialized to persist in the CF lung [68].

Gene loss is a general mechanism of bacterial evolution that avoids genetic redundancy and helps maintain a reasonable genome size in bacteria. Genome reduction is related to close association with host cells as observed for intracellular bacteria. Consequently, genomic reduction is a main force behind the evolution of parasitic and/or intracellular bacteria and the emergence of strict pathogens from OBPs [61,70,71]. A recent study compiling current genomics data confirmed that the evolution of specialized bacteria, including pathogenic bacteria, consists mainly of gene losses [72]. Moreover, extreme genome decay is often accompanied by a low GC% content [73]. A bacterium that establishes intracellular relationships is surrounded by the host cell’s metabolic products. Some bacterial metabolic pathways, redundant with those of the host cell, become non-essential and will be subject to genetic degradation due to the loss of selective pressure that acts upon them [74]. The closer the relationships the more genetic decay occurs. For example, the obligatory intracellular bacteria of the genus Rickettsia no longer have genes for amino acid and nucleotide biosynthesis. In the Alphaproteobacteria super-class, genome comparison of the facultative intracellular bacteria Bartonella henselae and rickettsiae shows that obligatory parasites display about 1,000 genes less than facultative ones. This repertoire of 1,000 genes probably represents the differences in metabolic needs between facultative and obligatory intracellular bacteria [75]. To take a more extreme example of endosymbiotism, the genome of a mitochondrion, which is an ancient alphaproteobacterium, displays about 800 genes less than Rickettsia prowazekii [74]. The genomic decay in intracellular bacteria is not only related to the change in selective pressure when the bacterium is associated with a narrow niche, but also to a change in population structure. Narrow niches-associated bacteria encounter a population bottleneck that leads to the emergence of small clonal populations isolated from foreign genetic information. In these conditions, slightly deleterious mutations accumulate without being compensated for by transfer of new genes [76]. The most radical mechanism of genomic reduction observed during adaptation to a narrow niche is loss of plasmids. Indeed, plasmids can impair bacterial fitness particularly by slowing down growth rate in the absence of environmental selective pressure [77]. For instance, none of the animal cell-associated alphaproteobacteria display plasmids while alphaproteobacteria living in soil or in the rhizosphere such as Rhizobiaceae carry various plasmids and often megaplasmids of more than 1 Mb [61].

Besides loss of plasmids, degradation and deletion of coding sequences in the chromosome also occur. Genomics provides examples of gradual degradation of gene contents with intermediates ranging from intact ORFs to complete gene disappearance through transcribed split ORFs, further split ORFs no longer transcribed and fully decayed but still recognizable ORFs [71]. All these intermediates are recognized as genomic scars, named pseudogenes. The gradual gene degradation is particularly obvious in comparative genomics of rickettsiae [71] but also occurred in the genome of other human pathogens such as Mycobacterium leprae, S. enterica serovar Typhi and Y. pestis [61]. More generally, the presence of a large number of pseudogenes is significant of genome reduction related to niche specialization. Large deletions due to accidents during recombination between repeated sequences are another mechanism of genome reduction [78]. Such deletions are detected in comparative genomics by the loss of a block of genes located between two repeated sequence copies as well as by the loss or the disturbance of one repeated sequence copy [76,78].

Comparative genomics in the super-class Alphaproteobacteria suggest that Rickettsiaceae lost 2,135 genes during their evolution to an intracellular lifestyle [79]. The comparison of 11 Rickettsia spp genomes shows that the differences in gene repertoires are mainly the result of differential gene losses from the rickettsial ancestor. The different genomic repertoires seem to play important roles in the adaptation of Rickettsia spp. to their various hosts, greater gene loss and subsequently smaller genomes being related to restricted host range. This diversity also appears to be crucial for the emergence of new species [79].

A longstanding belief was that ‘true’ pathogens are highly adapted to their hosts. The current vision shows pathogens as highly specialized in virulent behavior against their hosts. Georgiades et al. [54] recently proposed a scenario where pathogenicity of ‘true’ pathogens appears as an ‘unadapted’ behavior. Non-specialized bacteria such as OBPs enjoy a community lifestyle, which allows them to exchange genes resulting in an increased genome complexity whereas specialization results from a purifying selection [80]. At some point in their evolutionary history, the non-specialized bacteria can become specialized organisms in different niches. Subsequently, gene exchanges decrease and the gene repertoires undergo changes by differential reduction. The specialization of organisms results in gene loss. Regulatory gene loss and the resulting deregulation eventually leads to uncontrolled multiplication that destroys the ecosystem, i.e., host cells and tissues. In this evolutionary dead end scenario, the specialized bacteria appears as ‘unadapted’ [81].

We showed that pre-adaptation of environmental bacteria to pathogenicity for humans is related to genomic traits, while emergence of OBPs is the result of genome dynamics. The genera Pseudomonas, Aeromonas and Ochrobactrum will provide us with valuable examples of opportunistic behavior associated to particular genomic structure and evolution, with each of these OBPs displaying a refined association of genomic traits that shed light on their particular OBP behavior.

The core genome of P. aeruginosa strains from both clinical and environmental sources represents about 90% of the total genome [25,91]. Comparative genomic analysis of PAO1, PA14, PACS2, C3719 and PA2192 revealed that 5,021 genes are shared by all five genomes. More than 90% of these genes display a low average of nucleotide divergence with at least 98% identity in sequence [100]. As for most bacteria, core genome encodes housekeeping proteins involved in central metabolism. Surprisingly, pathogenic factors are encoded by the core genome of P. aeruginosa, i.e., shared by all strains irrespective of their origin. For example, a microarray-based comparison of 18 different P. aeruginosa strains found that 97% of the 267 examined PAO1 virulence-related genes were conserved across all strains [105]. A set of 980 genes, classically considered as non-essential because they are not involved in regulation, motility and virulence, shows no sequence variation at all among 36 clinical isolates. This indicates that these genes encode factors of evolutionary importance for the lifestyle of this successful environmental bacterium and opportunistic pathogen [106]. At the phenotypic level, common traits between environmental and opportunistic lifestyles can be highlighted. For instance, the ability to form polysaccharide-encased, surface-attached communities, known as biofilms, and to resist protozoan predation by injecting cytotoxic effector through a T3SS into the cytosol of eukaryotic cells can support persistence of P. aeruginosa both in nature and in colonization/infection of human beings.

However within the species, the fact that genome size varies between 5.2 and 7 Mbp [107] suggests accessory genome variability. Plasticity of accessory genome in P. aeruginosa is responsible for intra- and inter-clonal genome diversities that are easily detected by macrorestriction and PFGE. Several strains sharing the same genotype by single nucleotide polymorphism analysis in housekeeping genes were found to have different macrorestriction profiles. This fact supported the hypothesis that changes in P. aeruginosa genomes occur at a higher rate in the accessory DNA segments than in the conserved core genome [59]. Consequently, macrorestriction is the more discriminative method to trace clonal strains, for instance during outbreaks [60,108].

The elements of the accessory genome appear as “foreign blocks,” acquired by HGTs from other species or genera. The acquired blocks are interspersed among the genes of the core genome. Therefore, the overall architecture of the P. aeruginosa chromosome is often described as a mosaic structure. The fluidity of the mosaic observed by comparative genomics suggested recombination events but some physically distant loci exist in fixed combinations of genotypes, suggesting that the free flow of genes did not occur at all loci of the P. aeruginosa genome [107].

The accessory genome is not randomly scattered throughout the core genome but formed by dispersed polymorphic strain-specific segments, flanked by conserved genes used as anchors. The elements of the accessory genome can be present in subgroups of the P. aeruginosa population or only in single strains [104,107,109]. Mathee et al. [100] defined strain-specific regions of genomic plasticity (RGPs) as blocks of at least four contiguous open reading frames (ORFs) that are not conserved across all five P. aeruginosa genomes analyzed. Within the core genome, these RGPs are specific loci that act as hotspots for the insertion of accessory genes. In particular, tRNA genes in the core genome are frequently targeted for the insertion of accessory genetic elements [110]. Together with plasmids, the RGPs form a major part of the accessory genome. RGPs often contain mobile DNA elements acquired and kept by the host strain, such as integrative and conjugative elements (ICEs), phages, transposons or insertion sequence elements (IS). A recent review by Kung et al. [91] describes the different types of accessory elements. Interestingly, some RGPs, named replacement elements, belong to the core genome. They correspond to loci encoding proteins such as O-antigens, pilin, pyoverdin, that display less than 70% identity between homologs, despite being shared by all strains in the species. These replacement elements are under diversifying selection [104].

The analysis of seven sequenced P. aeruginosa strains (PAO1, PA14, LESB58, PA7, PA2192, C3719, PACS2) reveals 79 distinct RGPs [104]. The comparison of RGP gene content among strains reveals the highly customized nature of the genomes of 5 pathogenic P. aeruginosa [100]. RGPs appear to be the source of specialized functions that probably allow the survival of a particular strain in its environmental niche. For instance, in strain PA2192, evolution of RGP29 was linked to the acquisition of new metabolic capacities, such as growth in environments rich in abietane diterpenoid resins that are produced by conifers. Besides this outstanding behavior, PA2192 also establishes persistent infections in CF patients [100]. Acquisition of new traits appears as an expansion of the repertoire of functions rather than a specialization; indeed new traits do not eliminate pre-existing ones.

Studies at the population level show that the spread of dominant clones in both clinical and environmental habitats is a general feature of P. aeruginosa [107]. In Wiehlmann’s study [107], genotyping of a collection of 240 strains isolated from diverse habitats and geographical origins revealed that 16 most common clones made up half of the strain panel and were found to be widespread in multiple habitats and locations. Major clones are just as versatile in their local habitat as the whole P. aeruginosa population. For instance, Pirnay et al. [111] demonstrated that the local P. aeruginosa community in a Belgian river was almost as diverse as the entire terrestrial population of P. aeruginosa. The Belgian river harbored clones that also belonged to a collection of 73 clinical and environmental isolates previously collected across the world [112]. Such large-scale population studies did not detect strains specialized to particular niches such as human hosts. However, some specialized populations have been described, particularly for P. aeruginosa strains associated to the respiratory tract of CF patients.

P. aeruginosa infections in CF are a paradigm of how environmental bacteria can conquer, adapt to and persist in an atypical habitat and successfully evade defense mechanisms and chemotherapy in a susceptible host [113]. Considering core genome based phylogeny, the strain LESB58 hypervirulent for CF patients is the closest relative to the rhizosphere strain M18 [25]. However, dominant epidemic clones have been associated to CF. They display a phenotype distinct from other P. aeruginosa isolates characterized by reduced levels of virulence factor secretion [114], O-antigen deficiency [115], changes in LPS fatty acylation [116], amino acid auxotrophy [117], non-flagellation [118] and mutator phenotype [119]. The production of alginate and the formation of biofilms within the lung habitat are increased [120]. The genetic basis for this outstanding phenotype is point mutations in structural or regulatory genes, like mutS [121], mucA [122], rpoN [123] and lasR [124].

In early stages of infection in CF patients, the population of P. aeruginosa is diverse and displays phenotypes comparable to those of environmental isolates [125]. In contrast, adapted dominant epidemic strains are often identified from patients chronically infected with P. aeruginosa [126,127]. However, the dominant clones are probably present in the diverse population at the early stages of the disease [127]. The microevolution of P. aeruginosa populations in CF airways could be driven by hypermutable strains [119]. Proteomic and transcriptomic analysis of hypermutator CF strains showed an expression of genes involved in microaerobic lifestyle created by mucus in the lungs of CFpatients [128]. Besides point mutations, specialization for CF airways is accompanied by reorganization of the P. aeruginosa genomic structure by acquisition or loss of variable genomic regions [129] and genome rearrangements such as highly frequent large chromosomal inversions [130].

The genome organization and functionality of P. aeruginosa support the Swiss-army knife metaphor. Genome fluidity, acquisition of new specific genes (such as genes allowing life with toxic resins in ecosystems with conifers) and expression of phenotypic traits (such as alginate production in CF patient airways) account for adaptation to very particular niches, while mosaicism of core and accessory genes conserve versatile traits ensuring life under all conditions.

Aeromonads are found primarily in freshwater where virtually all species of the genus may be recovered. They have a sympatric lifestyle that favors HGTs and high genetic diversity [54]. Water is the most frequent source of human infection, (i) either for SSTIs, mainly caused by A. hydrophila and A. veronii, usually at a rare frequency, except in case of natural disasters when water-borne SSTIs become explosively frequent (e.g., hurricane Katrina in New Orleans in 2005, tsunami in Thailand in 2004, earthquake in Sichuan in 2008; [131]); (ii) or for digestive asymptomatic carriers and gastroenteritis, mostly associated with A. caviae and A. veronii [133], A. caviae being particularly associated with the environment of human beings since its density is greater in wastewater inflow than in outflow [134]; (iii) or for bacteremia, mainly caused by A. caviae and A. veronii, and indirectly associated with water through the gut from where these species originates most frequently.

Apart from water, aeromonads have been recovered from various types of ‘nurseries’ listed above, namely protozoa (amoeba), insects (mosquitoes) and worms (leeches), although the first two have not been studied in detail [135,136,137].

The genome of virulent A. hydrophila ATCC 7966^T (NC_008570) reveals an exceptionally versatile organism with a large potential for pathogenic processes and persistence in aquatic environments. Its genome contains a single circular 4.7 Mb chromosome with 61.5% G+C content and encodes 5,195 predicted coding sequences, 10 ribosomal-gene operons and 128 tRNA that may be correlated with an ability to rapidly respond to changing environmental conditions [132]. Numerous extracellular components are encoded, proportional to its genome size, and designed to respond rapidly to environmental fluctuations (e.g., two-component signal proteins, methyl-accepting chemotaxis proteins), suggesting that, from a pathogenic point of view, the bacterium is critically concerned with sensing the external environment and that it may be involved in, sensing and responding to signals originating in the host environment. Its rRNA operon evolution includes at least in part a mosaic evolution by HGT of partial rRNA operon fragments [138,139]. The genome is characterized by comprehensive metabolic abilities. Besides complete primary metabolism pathways, secondary metabolism is very versatile including many chitinolytic activities, substantial formate catabolism, several enzymes for altering toxic compounds, phosphonates, xenobiotics, plastics, dyes or nitroaromatic compounds. The organism also seems capable of tolerance to heavy metals and stress response and detoxification functions with superoxide dismutases, catalase, arsenate reductase, etc. [132]. Transporter genes are numerous and compare with those of pseudomonads and vibrios.

Few integrated regions have been observed, one of which includes lipopolysaccharide and carbohydrate synthesis that may be involved in chlorine resistance and the ability to form biofilm. Another one is a type 1 fimbriae gene cluster that may be involved in interaction with the host. Surprisingly, no transposase, resolvase or insertion sequence elements were found in A. hydrophila ATCC 7966^T genome, although these elements are usually associated with quick adaptation of free-living organisms.

The A. hydrophila ATCC 7966^T genome comprises a large panel of genes encoding virulence factors, including adhesion (type IV fimbriae and pili), toxins, iron acquisition, polar flagella, quorum-sensing regulation activators and antibiotic resistance among many others: all key processes for surviving in aquatic environments. Of note, genes encoding T3SS and one of its effectors (AexT) are absent although the role of the T3SS has been clearly established in the virulence of A. hydrophila. Since A. hydrophila ATCC 7966^T is one of the most virulent strains, it was suggested that the T3SS may be borne on a plasmid that may have been lost or that the absence of the T3SS is compensated by the presence of a type 6 secretion system or by the flagellar secretion apparatus, as described in Y. enterocolitica [132].

A. salmonicida differs from other Aeromonas species by being psychrophilic, non-motile and having a pathogenic spectrum limited to fish. A. salmonicida appears to be an example of the evolution toward pathogen specificity for a particular host within a group of mainly opportunistic pathogens or commensal bacteria. The A. salmonicida A449 genome (NC_009348) consists of a single circular 4.7 Mb chromosome with a G+C% of 58.5 and 4388 genes encoding 4086 proteins, 9 rRNA operons, and 110 tRNA genes. There are additionally 2 large and 3 small plasmids. Compared to A. hydrophila ATCC 7966^T, the A. salmonicida genome encodes a similar number of proteins although there has been gene gain or loss leading to a 9% difference in gene content [140,141]. This difference is characterized by a large number of mobile genetic elements (e.g., 88 IS) and of pseudogenes (170 compared to only 7 pseudogenes in A. hydrophila ATCC 7966^T) in the A. salmonicida A449 genome and 5 plasmids, totalizing 0.34 Mb in genome size. A substantial number of IS acquisitions resulted in gene disruptions, some of them leading to the loss of cell surface structures such as flagella or pili, and of enzymatic activities [73,140]. Additionally, the A. salmonicida A449 genome differs by its global genome inversion around the origin of replication consecutive to large rearrangements and the acquisition of a large sequence absent from the A. hydrophila ATCC 7966^T genome, and probably acquired by transfer since the sequence is bounded by transposons (Figure 1A). Beside and inside the inverted region, the genomic synteny is, on the whole, respected (Figure 1A).

Virulence factors include T3SS, a key factor for A. salmonicida virulence, and T6SS, although the latter seems to be not functional due to a key gene disruption. For both systems, genes are encoded either on the large plasmids or on the chromosome. The genome also contains genes encoding a secreted enzyme (e.g., collagenase), T2SS, iron acquisition, quorum sensing, (e.g., luxR) and pore-forming toxins (e.g., aerolysin). Interestingly, the A449 genome codes for an endotoxin 61% similar to the Bacillus thuringiensis insecticidal toxin CriET29, whose role remains unknown in A. salmonicida. Genes encoding adhesins like flagella, pili and surface layer are present, including the surface layer protein VapA, whose gene displays a low G+C% [141]. Attenuation of A. salmonicida by growth at high temperature is a result of IS insertion into the vapA gene [140]. Finally, genes of lateral and polar flagellae are present in the genome, but some of them are disrupted (e.g., lafA, flrA), which is consistent with the non-motile characteristic of the bacterium.

A. salmonicida evolved from a variable ancestral population to a rather genetically uniform host-adapted pathogen through accumulation of pseudogenes and IS, HGTs, plasmids and rearrangements [73].

The medicinal leech, Hirudo verbana, houses a remarkably simple two-member microbial community in the crop, (a pouch in its gut where ingested blood is stored): A. veronii and an uncultured Rikenella-like bacterium (Bacteroidetes). To date, only one other Aeromonas species, namely Aeromonas jandaei, has been recovered from the gut of a North American leech. The symbionts are thought to aid in digesting ingested blood, preventing other bacteria from colonizing the gut and providing essential nutrients. The 2 bacteria are organized in mixed-species polysaccharide embedded microcolonies that might form a biofilm associated with erythrocytes or that might float [135]. The complement system of the ingested blood remains active inside the crop and contributes to the specificity of the microbiota, A. veronii being a complement-resistant organism through its lipopolysaccharide layer. However, this factor is not enough to explain the symbiotic relationship because other complement resistant bacteria (e.g., P. aeruginosa, Staphylococcus aureus) have a markedly reduced ability to grow in the crop. Based on a signature-tagged mutagenesis, several genes have been involved in the interaction with the leech for a successful symbiotic relationship, including surface-expressed genes (e.g., lipopolysaccharide glycosylated surface proteins, exopolysaccharide formation, Braun’s major outer membrane protein), regulatory genes (e.g., regulating gene expression, enzymatic activity, protein synthesis rate), nutritional genes (e.g., threonine/serine transporter), host-interaction functions (e.g., T3S) or still unknown function genes [142,143]. In particular, the importance of regulatory genes for colonization shows the ability of A. veronii to recognize its host environment and regulate genes accordingly. Interestingly, the T3SS, also considered as a virulence factor [144,145], is essential for symbiosis with the leech and was detected in all A. veronii strains analyzed [146], except in the strain used for complete genome sequencing. It protects A. veronii against phagocytosis by leech hemocytes that circulate through the intraluminal fluid inside the crop, in order to maintain the symbiosis [147]. More generally, A. veronii seems to have an innate ability to colonize the gastrointestinal tract of blood feeding organisms (e.g., mosquitoes, vampire bats, leeches) and comprehensively to infect digestive tracts of multiple species with manifestations of interaction ranging from pathogenesis to mutualism [135,147].

Compared to the A. hydrophila ATCC 7966^T and A. salmonicida A449 genomes, A. veronii B565 (NC_015424) shows similar features with a single circular 4.55 Mb chromosome, a 58.7% G+C content, 4,057 protein coding genes, 10 rRNA operons and 102 tRNA genes [148]. Yet, several hundred genes present in A. hydrophila ATCC 7966^T and/or A. salmonicida A449 genomes were absent in A. veronii B565 genome (346, 329 and 366 genes, respectively), some of which are associated with mobile genetic elements. For example, 53% of the genes involved in interaction with the host were absent from A. veronii genome compared to A. hydrophila [142]. In addition, recombination events are frequent in A. veronii even in the T3SS genes. To date, no extensive genomic study, including comparative genomics, has been conducted on the A. veronii genome characteristics. We performed the alignment of A. veronii genome with those of A. hydrophila and A. salmonicida using MAUVE software (Figure 1B,C). In both comparisons, we observed extensive rearrangements with a loss of gene synteny. Considering the global genome conservation between A. hydrophila and A. salmonicida, this result suggested that A. veronii encountered a diversification process, which could be related to a variety of niches and hosts. The type of genes involved in A. veronii mutualistic strains is similar to those found in pathogenic strains [144,147]. Although it remains unclear why the same function acts as colonizing in invertebrates and pathogenic in human or fish, the leech colonization may be a good ‘training course’ for A. veronii in pre-adaptation to digestive tract colonization, resistance to complement and serum lysis and resistance to phagocytosis. These qualities may in turn be used to sustain human infection, especially for gastroenteritis and bacteremia.

The complete genome of O. anthropi ATCC 49188^T has been recently but briefly reported [150]. Similar to other Rhizobiales [151] and to other members of the genus Ochrobactrum [57], the genome of O. anthropi ATCC 49188^T consists of multiple circular chromosomes, a replicon of about 2.9 Mb with the prototypical bacterial chromosome origin of replication (oriC), and a second replicon of about 1.9 Mb with a repABC origin commonly found in other secondary chromosomes and plasmids of the Rhizobiales [150]. In addition, the O. anthropi ATCC 49188^T genome contains four plasmids carrying genes related to stabilization factors, including genes encoding ParB-like, PilT-type and toxin-antitoxin addiction systems. Plasmids include a large number of genes encoding transposases, integrases and several transporters, suggesting that they contribute to bacterial fitness via genes acquired by HGT [150]. Genome size appears to be exceptionally variable (from 5.06 to 8.12 Mb) in the species O. anthropi, variations bearing mostly on number and size of plasmids [57]. The strain-specific plasmid content stresses the role of plasmids for adaptation of each strain to a particular niche, therefore supporting versatility in the species O. anthropi.

The draft sequence of O. intermedium is now available, allowing a preliminary comparison of Brucella suis, O. anthropi and O. intermedium genomes. The alignment performed with the MAUVE software is shown in Figure 2 and Figure 3. The large chromosomes (chr I) of the 3 species are similar and roughly colinear but display clear differences in size related to indels scattered in the sequences (Figure 2). Compared to O. anthropi, chr I of O. intermedium is very similar in content but is smaller by about 300 kb, suggesting a reductive process confirmed by a higher % of pseudogenes than observed in O. anthropi. Moreover, O. intermedium compared to O. anthropi presents a large inversion involving two thirds of the genome. Surprisingly, colinearity is respected between O. intermedium and B. suis genomes, the main differences between them corresponding to deletions leading to a genome size smaller by about 700 kb for B. suis (Figure 2). This confirms the previously described genome degradation in brucellae [152]. The major deletion observed by MAUVE comparison corresponds to a large fragment of about 430 kb, present in the Ochrobactrum but not the Brucella chr I. About half of the corresponding genes are absent from Brucella spp. genomes and another half are inserted in chr II of B. suis. Surprisingly, some of the latter genes are inserted in chr II of B. suis ATCC 23445 and in chr I of B. suis 1330. This result confirms that recombination events occurred between the two chromosomes in brucellae genomes as previously described for recombination between rrn operons [78]. Comparison of the small chromosomes (chr II) of B. suis, O. anthropi and O. intermedium genomes (Figure 3) shows more indel events and lack of synteny than observed for chr I. Again. the chr II of B. suis is smaller than those of O. anthropi and O. intermedium. Despite indels and rearrangements, chr II of O. anthropi and O. intermedium remain globally colinear while B. suis displays an extensively modified chr II. Clearly, chr II supports the genomic diversity in Brucellaceae and probably ensures the adaptation to diverse niches.

Plasmid POAN01 of O. anthropi includes a complete set of type IV secretion system genes (operon virB) whereas this operon is found on the second chromosome of O. intermedium. This suggests the transfer of virulence traits from a disposable to an essential replicon leading to immortalization of the function.

Although previous studies have supported the notion of Brucella as a monospecies genus, it is now accepted that the genus Brucella is divided into ten species, named according to their preferential hosts, except for Brucella inopinata, for which a natural host remains unknown. Genome sequencing of Brucella melitensis, and B. suis, the most pathogenic among brucellae, demonstrated a high level of similarity between the two genomes, with more than 90% of genes having more than 98% nucleotide identity [153]. It has been proposed that the unique complement of pseudogenes in each of the Brucella species may contribute to their differing degrees of infectivity and host preference [154]. This suggested that the differential genome degradation among brucellae confers their specific pathogenic behavior. Compared to B. suis and B. melitensis, the biology of Brucella ovis, which is non-pathogenic for humans, but closely associated to and pathogenic for sheep, appears to be in part the result of genome degradation [152], as suggested by the % of pseudogenes (Table 3). This genome reduction correlates with the adaptation to a narrower niche in a unique host. Moreover, the majority of human brucellosis cases occur via ingestion of contaminated dairy products, except for B. ovis. This suggests that this species has lost the ability to infect via the oral route in relation with genome degradation, particularly the loss of urease that is required for survival of stomach acidity by Brucella spp. [155]. By contrast, the genome of the OBP O. anthropi genome displays a very low % of pseudogenes, evoking the conservation of a large repertoire adapted to life in versatile niches.

In a previous study, we analyzed clonal strains chronologically isolated from respiratory tract samples of a patient with chronic O. intermedium carriage over a one-year period. We observed the loss of one rrn copy and a large genomic deletion that occurred between the second (AV2) and the third isolate (ADV3) obtained from the patient over a period of one month. In vitro, variations in colony aspect and growth rate were observed between ADV2 and ADV3. The new genomic organization and phenotype were maintained in four subsequent isolates obtained over a 4-month period. This suggests that the new genomic organization gave selective advantage to the strain and that bacterial life in a very narrow ecological niche, in this case the human respiratory tract, led to a reductive evolution process [90]. The rearrangement consisted in a 150-kb deletion corresponding to the whole genomic region comprised between the two rrn copies of the small chromosome. The deletion occurred by homologous recombination between the 2 rrn copies and resulted in the loss of one copy. Interestingly, the genomic skeleton obtained after the deletion event occurred evoked the genomic structure of Brucella spp. with reduced genomic size and only 3 rrn operons.

This report of genomic reduction in real time, taken together with the preliminary results concerning the total sequence of O. intermedium i.e., reduction of chr I, % of pseudogenes and ‘immortalization’ of virB on chr I, suggest that this bacterium might be involved in a similar specialization process as that encountered by Brucella in its own evolution. Data at the population level confirmed that most strains of O. intermedium are isolated from human beings (Aujoulat, personal communication).